Research And Implementation Of Emotional Dictionary Based On Sentence Pattern

In today’s era of network development, e-commerce is on the rise. Every industries has a store in the electronic except a physical store. The product reviews in the e-commerce platform has much important information. Manufacturers can use the imformation to improve their products and develop their own strengths. How to identify and extract the information and classified information which is a hot topic on the Internet network comments mining.By studying the existing techniques of building emotion dictionary. The paper summarized the advantages and disadvantages of the other methods.I propose a construction method based on cross-cutting emotion sentence dictionary based on the dictionary constructed to build emotion dictionary based on setence pattern according to the technical requirements. On the basis of showing the basic concepts of opinion mining elaborate, the paper introduce the technics related on emotion dictionnary in detail and analysising other technology of building emotion dictionary. The paper use sentence pattern to biuld the dictionary. Through the study of comments sentences, finding out comments sentence characteristics of each sentence. And calculated the emotional value ofunknown emotion word to generate emotion dictionary based on these characteristics. The advantages of the method to build emotion dictionary p roposed:1. The emotion dictionary generated is not always exitst.but generated by he operation of the system according to the input corpus review. That is, to calculate the corresponding emotional value by algorithm according the input corpus.2. The emotion dictionary generated in different areas is unrelated. Each field must be corresponding emotion in the field dictionary. Not only the word of emotion dictionarys can be different, but also emotion word of one dictinary could have different emtional value.3. The word of emotion dictionary generated is no exiat alone. Emotional value of each emotion word is attached to corresponding features. The emotinal value of each emotiona word on a different feature is not the same. So the same word in different emotional characteristics, which may not be the same part of speech the word emotion.Finally, according to the division of functional modules to test the system, get the accuracy and precision of the method, experiments show that the system basically meet the design requirements.